trainer_state.json

{
  "best_metric": 0.8005333333333333,
  "best_model_checkpoint": "./resnet_finetuned_models_dataset/CIFAR100/50_from_100/microsoft_resnet-101/model_idx_0164/checkpoints/checkpoint-2664",
  "epoch": 8.0,
  "eval_steps": 500,
  "global_step": 2664,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "grad_norm": 283.3953857421875,
      "learning_rate": 2.9095389311788626e-05,
      "loss": 3.86,
      "step": 333
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.32853333333333334,
      "eval_loss": 3.766831159591675,
      "eval_runtime": 15.7132,
      "eval_samples_per_second": 238.653,
      "eval_steps_per_second": 3.755,
      "step": 333
    },
    {
      "epoch": 2.0,
      "grad_norm": 226.50132751464844,
      "learning_rate": 2.649066664678467e-05,
      "loss": 3.3112,
      "step": 666
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.5541333333333334,
      "eval_loss": 2.649690628051758,
      "eval_runtime": 15.5655,
      "eval_samples_per_second": 240.918,
      "eval_steps_per_second": 3.79,
      "step": 666
    },
    {
      "epoch": 3.0,
      "grad_norm": 118.48308563232422,
      "learning_rate": 2.25e-05,
      "loss": 1.9707,
      "step": 999
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.6914666666666667,
      "eval_loss": 1.383646845817566,
      "eval_runtime": 15.5288,
      "eval_samples_per_second": 241.487,
      "eval_steps_per_second": 3.799,
      "step": 999
    },
    {
      "epoch": 4.0,
      "grad_norm": 125.05719757080078,
      "learning_rate": 1.760472266500396e-05,
      "loss": 1.2026,
      "step": 1332
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.7378666666666667,
      "eval_loss": 1.000968337059021,
      "eval_runtime": 15.6208,
      "eval_samples_per_second": 240.065,
      "eval_steps_per_second": 3.777,
      "step": 1332
    },
    {
      "epoch": 5.0,
      "grad_norm": 94.04993438720703,
      "learning_rate": 1.2395277334996045e-05,
      "loss": 0.9423,
      "step": 1665
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.7821333333333333,
      "eval_loss": 0.8265097737312317,
      "eval_runtime": 15.5901,
      "eval_samples_per_second": 240.537,
      "eval_steps_per_second": 3.784,
      "step": 1665
    },
    {
      "epoch": 6.0,
      "grad_norm": 138.1405792236328,
      "learning_rate": 7.500000000000004e-06,
      "loss": 0.8229,
      "step": 1998
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.7829333333333334,
      "eval_loss": 0.7733106017112732,
      "eval_runtime": 15.7186,
      "eval_samples_per_second": 238.57,
      "eval_steps_per_second": 3.754,
      "step": 1998
    },
    {
      "epoch": 7.0,
      "grad_norm": 254.40843200683594,
      "learning_rate": 3.5093333532153316e-06,
      "loss": 0.769,
      "step": 2331
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.7957333333333333,
      "eval_loss": 0.7261528372764587,
      "eval_runtime": 15.4798,
      "eval_samples_per_second": 242.252,
      "eval_steps_per_second": 3.811,
      "step": 2331
    },
    {
      "epoch": 8.0,
      "grad_norm": 109.3260726928711,
      "learning_rate": 9.046106882113753e-07,
      "loss": 0.7434,
      "step": 2664
    },
    {
      "epoch": 8.0,
      "eval_accuracy": 0.8005333333333333,
      "eval_loss": 0.722422182559967,
      "eval_runtime": 15.6227,
      "eval_samples_per_second": 240.036,
      "eval_steps_per_second": 3.777,
      "step": 2664
    }
  ],
  "logging_steps": 500,
  "max_steps": 2997,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 9,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 6.5411433916416e+18,
  "train_batch_size": 64,
  "trial_name": null,
  "trial_params": null
}